Thinking in Systems

Thinking in Systems

Thinking In Systems lays out a bunch of different ways for thinking about 'systems'. Frankly, I think I prefer the slightly-more-applied versions that you find in Will Larson's books. But all the same, there are some good ideas here that make the book worth reading.

A system is formed not just of elements (physical things we can sense) but also of the relationship between those elements. When a new president is elected, the elements are all the same (all 300m people) but the rules and relationships governing the system have changed. Same thing for a basketball team. The elements are the players, but how they interact can change. IMPORTANT: you can both change the relationships and the components of the system.

Systems may be made of subsystems, fractals nested and each with their own purpose.

A useful way to think about systems is as a combination of stocks and flows. Stocks are physical quantities you can see and observe. Flows are the inputs and outputs to them. These can reach dynamic equilibrium where they are the same. In general, humans are better at observing stocks than flows, and inflows more than outflows. It's easier to say "let's identify more oil wells" than it is to figure out how to use less oil.

We naturally apply stocks anywhere that we need a fixed rate of something. Water reservoirs, oil tankers, storehouses for grain are all good examples.

Many stock and flow systems are feedback loops. If there's too much wheat in the market, farmers will plant something else that's higher value. As your bank account lowers, you will spend less or find ways to earn more.

Reinforcing feedback loops are super important to shaping behavior. These occur when the output increases the input (apropos right now with inflation and consumer spending). When A causes B to increase... also ask yourself... does B cause A?

There's always an information delay in any system where changes take time to propagate. Sounds obvious but it won't happen right away! If you have a long feedback loop (imagine your shower heating up after you change the nozzle), you'll tend to get frustrated unless you can look at leading indicators.

A few examples

  • single stock: thermostat. Stock is the desired temperature. Flows are furnace heat in, and leaking heat out.
  • Reinforcing + balancing loop: population. The stock is population. More growth leads to more reinforcement. Mortality leads to lower population. Fertility rate generally decreases as nations develop.

Reinforcing feedback loops

There are two primary kinds of reinforcing loops: stabilizers and amplifiers. Stabilizing loops will mute the result wrt the inputs: if you are feeling low-energy, you might have some coffee to increase your energy. If you are feeling jittery, you will probably reduce the amount of coffee.

Amplifying loops work in the opposite manner. They will continue to grow and grow and grow in response to input. A good example would be “less investment drives lower revenue drives less investment”.